Objective: To describe consultations provided by child abuse pediatricians for cases referred by child protective services (CPS); compare the opinions of the likelihood of child maltreatment of the initial physician, CPS, and the child abuse pediatrician; and examine predictors of the experts' opinions. Methods: Cases were referred by CPS for consultations between March 1, 1998, and June 30, 2005, to 2 child abuse pediatricians at Yale-New Haven Children's Hospital. We abstracted demographic and clinical information and the opinions of the initial physician, CPS, and the child abuse expert, each coded using a 5-point scale from definite maltreatment to definite benign cause (eg, accident). Results: Of 187 cases, 50.3% occurred in children younger than 1 year of age. Children's most serious injuries were fractures (50.8%), burns (16.6%), and bruises/abrasions (15.0%). The child abuse experts' opinions were 47.6% definite or probable maltreatment, 8.6% uncertain, and 43.9% definite or probable benign. Of the 119 cases with opinions from all 3 assessors, the expert agreed with the physician in 57.1% of cases (κ = 0.34) and with CPS in 64.7% (κ = 0.42). The best predictor of the expert's opinion that the injury was due to maltreatment was agreement between the physician and CPS that maltreatment had occurred. Conclusions: Levels of agreement were fair to poor between the child abuse expert and either the physician or CPS. Child abuse experts' opinions have important value in selected cases to confirm previous assessments by the physician and/or CPS, or to change the opinion of the case. Copyright © 2011 by Academic Pediatric Association.